PyNN on SpiNNaker Installation Guide

Follow me on GitHub

The version described here is no longer supported.

Home page for current version

This guide will detail how to install the release versions of the tools required to run simulations on SpiNNaker using the PyNN scripting language. This has been tested using Fedora Linux 20 64-bit, Ubuntu Linux 14.04 LTS 64-bit, Windows 8.1 64-bit, and Mac OS X Mavericks, but it should in principle work on 32-bit and 64-bit versions of Fedora and Ubuntu Linux, Windows 7 and 8, and Mac OS X.

Contents

Ubuntu Linux Requirements

  1. Install Python
    sudo apt-get install python2.7
  2. Install numpy
    sudo apt-get install python-numpy
  3. Install scipy
    sudo apt-get install python-scipy
  4. Install pip
    sudo apt-get install python-pip
  5. Optional: Install matplotlib (often used in PyNN scripts for plotting graphs)
    sudo apt-get install python-matplotlib

Continue to the Standard Central Installation, User-only Installation or Virtualenv Installation to install the remaining requirements.

Fedora Linux Requirements

  1. Install Python
    sudo yum install python
  2. Install numpy
    sudo yum install numpy
  3. Install scipy
    sudo yum install scipy
  4. Install pip
    sudo yum install python-pip
  5. Optional: Install matplotlib (often used in PyNN scripts for plotting graphs)
    sudo yum install python-matplotlib

Continue to the Standard Central Installation, User-only Installation or Virtualenv Installation to install the remaining requirements.

Mac OS X Requirements

  1. Download and install MacPorts
  2. Install python
    sudo port install python27
    sudo port select --set python python27
  3. Install numpy
    sudo port install py27-numpy
  4. Install scipy
    sudo port install py27-scipy
  5. Install pip
    sudo port install py27-pip
    sudo port select --set pip pip27
  6. Optional: Install matplotlib (often used in PyNN scripts for plotting graphs)
    sudo port install py27-matplotlib

Continue to the Standard Central Installation, User-only Installation or Virtualenv Installation to install the remaining requirements.

32-bit Windows Requirements

  1. Download and install Python
  2. Download and install numpy
  3. Download and install scipy
  4. Download and install setuptools and [pip] (https://github.com/SpiNNakerManchester/github.SpiNNakerManchester.io/releases/download/v1.0-win32/pip-1.5.6.win32-py2.7.exe)
  5. Optional: Download and install matplotlib (often used in PyNN scripts for plotting graphs)
    1. Download and install dateutil, pyparsing, six and matplotlib

Continue to edit your PATH, before installing the remaining requirements.

64-bit Windows Requirements

  1. Download and install Python
  2. Download and install numpy
  3. Download and install scipy
  4. Download and install setuptools and [pip] (https://github.com/SpiNNakerManchester/github.SpiNNakerManchester.io/releases/download/v1.0-win64/pip-1.5.6.win-amd64-py2.7.exe)
  5. Optional: Download and install matplotlib (often used in PyNN scripts for plotting graphs)
    1. Download and install dateutil, pyparsing, six and matplotlib

Continue to edit your PATH, before installing the remaining requirements.

Windows Path Settings

  1. Edit your PATH environment variable to include the Python27, Python27\Scripts and (if installed) gtk\bin directories.
    1. Go to “Control Panel”
    2. Go to “System” (may be under “System and Security”)
    3. Select “Advanced System Settings”
    4. Select “Environment Variables”
    5. Select the “Path” in the “System Variables” box at the bottom
    6. Click on “Edit”
    7. Add the new directory by putting it before the existing string, followed by a single semi-colon; if you installed the dependencies to the default/recommended locations, prepend Path with:
      C:\Python27;C:\Python27\Scripts;C:\gtk\bin;

Continue to the Standard Central Installation, User-only Installation or Virtualenv Installation to install the remaining requirements.

Standard Central Installation

These instructions will install the required packages in a central location. If you are installing on Linux, you must have root access to do this (or prepend each command with sudo), and Windows users should run these commands from a command prompt run as Administrator (right-click the shortcut for the command prompt and select “Run as administrator”).

  1. install spinnMachine pip install SpiNNMachine=2015.002
  2. install pacman pip install spinnaker_pacman=2015.002
  3. install spinnman pip install spinnman=2015.002
  4. install data specification pip install Spinnaker_DataSpecification=2015.002
  5. Install sPyNNaker
    pip install sPyNNaker=2015.003
  6. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

User-only Installation

These instructions will install the required packages only for the current user (in your home directory). This can avoid issues where you don’t have root access to the machine you are installing on, or on a shared machine where the dependencies might otherwise conflict with those of other users.

  1. install spinnMachine pip install SpiNNMachine=2015.002 --user
  2. install pacman pip install spinnaker_pacman=2015.002 --user
  3. install spinnman pip install spinnman=2015.002 --user
  4. install data specification pip install Spinnaker_DataSpecification=2015.002 --user
  5. Install sPyNNaker
    pip install sPyNNaker=2015.003 --user
  6. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003 --user

You can now configure your environment.

Virtualenv Installation

These instructions will install the required packages only in a virtualenv. Like the user-only installation, this can help when you don’t have root access or are on a shared machine. Additionally, it will help when you have several packages with conflicting dependencies, or those that occupy the same namespace (such as pyNN.spiNNaker if you have an older version of the toolchain).

The installation of virtualenv and the linking to the external libraries is platform dependent.

Ubuntu Virtualenv Installation

  1. Install virtualenv
    sudo pip install virtualenv
  2. Create a virtualenv; in the following can be replaced by the name of your choice ```virtualenv ```
  3. Activate the virtualenv
    source <name>/bin/activate
  4. Link numpy to the virtualenv
    ln -s /usr/lib/python2.7/dist-packages/numpy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  5. Link scipy to the virtualenv
    ln -s /usr/lib/python2.7/dist-packages/scipy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  6. Optional: link matplotlib to the virtualenv and install dependencies (only if you chose to install matplotlib)
    pip install python-dateutil
    pip install pyparsing
    pip install six
    ln -s /usr/lib/pymodules/python2.7/matplotlib $VIRTUAL_ENV/lib/python2.7/site-packages/
    ln -s /usr/lib/pymodules/python2.7/pylab.py $VIRTUAL_ENV/lib/python2.7/site-packages/
  7. install spinnMachine pip install SpiNNMachine=2015.002
  8. install pacman pip install spinnaker_pacman=2015.002
  9. install spinnman pip install spinnman=2015.002
  10. install data specification pip install Spinnaker_DataSpecification=2015.002
  11. Install sPyNNaker
    pip install sPyNNaker=2015.003
  12. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

32-bit Fedora Virtualenv Installation

  1. Install virtualenv
    sudo pip install virtualenv
  2. Create a virtualenv; in the following can be replaced by the name of your choice ```virtualenv ```
  3. Activate the virtualenv
    source <name>/bin/activate
  4. Link numpy to the virtualenv
    ln -s /usr/lib/python2.7/site-packages/numpy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  5. Link scipy to the virtualenv
    ln -s /usr/lib/python2.7/site-packages/scipy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  6. Optional: link matplotlib to the virtualenv and install dependencies (only if you chose to install matplotlib)
    pip install python-dateutil
    pip install pyparsing
    pip install six
    ln -s /usr/lib/python2.7/site-packages/matplotlib $VIRTUAL_ENV/lib/python2.7/site-packages/
    ln -s /usr/lib/python2.7/site-packages/pylab.py $VIRTUAL_ENV/lib/python2.7/site-packages/
  7. install spinnMachine pip install SpiNNMachine=2015.002
  8. install pacman pip install spinnaker_pacman=2015.002
  9. install spinnman pip install spinnman=2015.002
  10. install data specification pip install Spinnaker_DataSpecification=2015.002
  11. Install sPyNNaker
    pip install sPyNNaker=2015.003
  12. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

64-bit Fedora Virtualenv Installation

  1. Install virtualenv
    sudo pip install virtualenv
  2. Create a virtualenv; in the following can be replaced by the name of your choice ```virtualenv ```
  3. Activate the virtualenv
    source <name>/bin/activate
  4. Link numpy to the virtualenv
    ln -s /usr/lib64/python2.7/site-packages/numpy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  5. Link scipy to the virtualenv
    ln -s /usr/lib64/python2.7/site-packages/scipy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  6. Optional: link matplotlib to the virtualenv and install dependencies (only if you chose to install matplotlib)
    pip install python-dateutil
    pip install pyparsing
    pip install six
    ln -s /usr/lib64/python2.7/site-packages/matplotlib $VIRTUAL_ENV/lib/python2.7/site-packages/
    ln -s /usr/lib64/python2.7/site-packages/pylab.py $VIRTUAL_ENV/lib/python2.7/site-packages/
  7. install spinnMachine pip install SpiNNMachine=2015.002
  8. install pacman pip install spinnaker_pacman=2015.002
  9. install spinnman pip install spinnman=2015.002
  10. install data specification pip install Spinnaker_DataSpecification=2015.002
  11. Install sPyNNaker
    pip install sPyNNaker=2015.003
  12. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

Mac OS X Virtualenv Installation

  1. Install virtualenv
    sudo pip install virtualenv
  2. Create a virtualenv; in the following can be replaced by the name of your choice ```virtualenv ```
  3. Activate the virtualenv
    source <name>/bin/activate
  4. Link numpy to the virtualenv
    ln -s /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  5. Link scipy to the virtualenv
    ln -s /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/scipy* $VIRTUAL_ENV/lib/python2.7/site-packages/
  6. Optional: link matplotlib to the virtualenv and install dependencies (only if you chose to install matplotlib)
    pip install python-dateutil
    pip install pyparsing
    pip install six
    ln -s /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib $VIRTUAL_ENV/lib/python2.7/site-packages/
    ln -s /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pylab.py $VIRTUAL_ENV/lib/python2.7/site-packages/
  7. install spinnMachine pip install SpiNNMachine=2015.002
  8. install pacman pip install spinnaker_pacman=2015.002
  9. install spinnman pip install spinnman=2015.002
  10. install data specification pip install Spinnaker_DataSpecification=2015.002
  11. Install sPyNNaker
    pip install sPyNNaker=2015.003
  12. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

Windows Virtualenv Installation

  1. Open a console as administrator and cd to your home directory
    cd %HOMEPATH%
  2. Install virtualenv
    pip install virtualenv
  3. Create a virtualenv; in the following can be replaced by the name of your choice ```virtualenv ```
  4. Activate the virtualenv
    <name>\Scripts\activate.bat
  5. Link numpy to the virtualenv
    mklink /D %VIRTUAL_ENV%\Lib\site-packages\numpy C:\Python27\Lib\site-packages\numpy
    mklink %VIRTUAL_ENV%\Lib\site-packages\numpy-1.9.1-py2.7.egg-info C:\Python27\site-packages\numpy-1.9.1-py2.7.egg-info
  6. Link scipy to the virtualenv
    mklink /D %VIRTUAL_ENV%\Lib\site-packages\scipy C:\Python27\Lib\site-packages\scipy
    mklink %VIRTUAL_ENV%\Lib\site-packages\scipy-0.14.1rc1-py2.7.egg-info C:\Python27\site-packages\scipy-0.14.1rc1-py2.7.egg-info
  7. Optional: link matplotlib to the virtualenv and install dependencies (only if you chose to install matplotlib)
    pip install python-dateutil
    pip install pyparsing
    pip install six
    mklink /D %VIRTUAL_ENV%\Lib\site-packages\matplotlib C:\Python27\Lib\site-packages\matplotlib
    mklink %VIRTUAL_ENV%\Lib\site-packages\pylab.py C:\Python27\Lib\site-packages\pylab.py
  8. Optional: To make matplotlib work within a virtualenv, create the following environment variables:
    TCL_LIBRARY: C:\Python27\tcl\tcl8.5
    TK_LIBRARY: C:\Python27\tcl\tk8.5
  9. install spinnMachine pip install SpiNNMachine=2015.002
  10. install pacman pip install spinnaker_pacman=2015.002
  11. install spinnman pip install spinnman=2015.002
  12. install data specification pip install Spinnaker_DataSpecification=2015.002
  13. Install sPyNNaker
    pip install sPyNNaker=2015.003
  14. Install pyNN-SpiNNaker
    pip install pyNN-SpiNNaker=2015.003

You can now configure your environment.

Configuration

When SpyNNaker is first called, if a configuration file is not found, it will create one in your home directory and exit. It is possible to ask SpyNNaker to do this before you run your first simulation as follows:
python -c "import pyNN.spiNNaker"

Note that if you have previously installed a version of the spiNNaker software, you may already have a file called “.pacman.cfg” in your home directory. In this case, SpyNNaker will attempt to use this file for its paramters. If you don’t have this file, a new file called “.spynnaker.cfg” will be created in your home directory. You must edit this file to ensure that SpyNNaker can access your SpiNNaker machine. Upon opening this file, the part to alter will look like the following:

[Machine]
machineName = None
version = None

Within the file, you should set machineName to the IP address or hostname of your SpiNNaker machine, and version to the version of your SpiNNaker board; this will almost certainly be “3” for a 4-chip board or “5” on a 48-chip board.

Running some examples

  1. Download the examples from here (zip) or here (tar.gz)
  2. Extract the archive
  3. Go to the “examples” folder
  4. Run:
    python va_benchmark.py
  5. You will see the system go through a series of processes from partitioning, to placement, to routing and finally to loading and running.
  6. Once the example has finished, you should see a graph, that will look something like this:
    VABenchmarkSpikes If you get the output above, you have successfully installed your system.

Trouble Shooting Section

  1. UnicodeDecodeError: if you experience the error UnicodeDecodeError: ‘ascii’ codec can’t decode byte 0xb0 in position 1: ordinal not in range(128) Then to solve it follow these instuctions:

  2. Edit C:\Python27\Lib\mimetypes.py and add these codes:

  if sys.getdefaultencoding() != 'gbk':
      reload(sys)
      sys.setdefaultencoding('gbk')

before the following line:

   default_encoding = sys.getdefaultencoding()